Rural and Small Town Canada Analysis Bulletin
Vol. 7, No. 5 (January 2009)
Catalogue no. 21-006-X
Domestic Water Use:
The Relevance of Rurality in Quantity Used and Perceived Quality
David Hardie and Alessandro Alasia, Statistics Canada
Highlights
•
•
•
•
•
One-third of rural and small town residents rely on private wells for their drinking water.
Rural residents connected to a municipal water system have a higher per capita use of water
than urban residents.
Water use appears to have a stronger association with economic incentives than with location
characteristics. Households in areas with a higher proportion of water meters use less water
than households in areas with a lower proportion of water metering.
For households using tap water for drinking, rural households are less likely to treat their
water than urban households.
Locational characteristics are significant factors in determining perception of water quality as
measured by the choice of treating tap water for domestic consumption. Water source
(municipal systems or private wells) does not seem to affect water quality perception.
Introduction
Water supply and management is a growing
concern for residents of rural and urban
municipalities in Canada. Over the past few years,
several cases of water contamination and an
intensified debate on environmental problems
have increased the attention of the general public
to water issues. In decision-making, as well as in
the public debate, there are two major dimensions
that come into play: growing water demand and
ensuring good water quality. With the expansion
of urban agglomerations, the management of this
natural resource is also becoming a new area of
potential tension and/or collaboration between
municipalities that typically have different
patterns of water utilization and management
systems, such as rural and urban municipalities.
Rural and Small Town Canada Analysis Bulletin, Vol. 7, No. 5
Rural and Small Town Canada
Analysis Bulletin
ISSN 1481-0964
ISBN 978-0-662-48111-9
Editor: Ray D. Bollman
Associate Editor: Neil Rothwell
Published in collaboration with The Rural Secretariat,
Agriculture and Agri-Food Canada. The Rural and
Small Town Canada Analysis Bulletin is an
occasional publication of the Agriculture Division of
Statistics Canada. To obtain a single issue, visit our
website at www.statcan.gc.ca and select "Publications."
Contact the Agriculture Division at:
Agriculture Division, Statistics Canada
Ottawa, Ontario K1A 0T6
Toll free telephone number: 1-800-465-1991
Internet: [email protected]
Fax: (613) 951-3868
Editorial Committee: Denis Chartrand, Jeffrey Smith,
Heather Clemenson, Bishnu Saha, Marco Morin,
Aurelie Mogan and Deb Harper.
Special thanks to Véronique Julien and Josée
Bourdeau for their contribution in the publication
process.
Published by authority of the Minister responsible for
Statistics Canada.
© Minister of Industry, 2009.
All rights reserved. The content of this electronic
publication may be reproduced, in whole or in part,
and by any means, without further permission from
Statistics Canada, subject to the following conditions:
that it be done solely for the purposes of private study,
research, criticism, review or newspaper summary,
and/or for non-commercial purposes; and that
Statistics Canada be fully acknowledged as follows:
Source (or “Adapted from”, if appropriate): Statistics
Canada, year of publication, name of product,
catalogue number, volume and issue numbers,
reference period and page(s). Otherwise, no part of
this publication may be reproduced, stored in a
retrieval system or transmitted in any form, by any
2
means–electronic, mechanical or photocopy–or for
any purposes without prior written permission of
Licensing Services, Client Services Division, Statistics
Canada, Ottawa, Ontario, Canada K1A 0T6.
Note of appreciation
Canada owes the success of its statistical system to a
long-standing partnership between Statistics Canada,
the citizens of Canada, its businesses, governments
and other institutions. Accurate and timely statistical
information could not be produced without their
continued cooperation and goodwill.
Standards of service to the public
Statistics Canada is committed to serving its clients in a
prompt, reliable and courteous manner. To this end,
Statistics Canada has developed standards of service
that its employees observe. To obtain a copy of these
service standards, please contact Statistics Canada toll
free at 1-800-263-1136. The service standards are also
published on www.statcan.gc.ca under “About us” >
“Providing services to Canadians”.
Symbols
The following standard symbols are used in
Statistics Canada publications:
.
..
...
0
0s
p
r
x
E
F
not available for any reference period
not available for a specific reference
period
not applicable
true zero or a value rounded to zero
value rounded to 0 (zero) where there
is a meaningful distinction between
true zero and the value that was
rounded
preliminary
revised
suppressed to meet the confidentiality
requirements of the Statistics Act
use with caution
too unreliable to be published
Statistics Canada – Catalogue no. 21-006-X
Rural and Small Town Canada Analysis Bulletin, Vol. 7, No. 5
Concerns about water scarcity have global and
long-term implications. However, in a local
perspective, as water is normally supplied by
municipal governments, “scarcity” assumes a
specific connotation. The constraint on water
supply is typically determined by the conditions
and capacity of municipal infrastructures. The
status of these infrastructures has become, or is
expected to become in the near future, a limiting
factor to further growth for some municipalities. It
is a dilemma that municipal governments have
noticed, and they have begun to take action to
manage water demand (Kingston Whig-Standard,
2006; Munro, 2004).
There is often a heated debate about which policy
focus, or which combination of measures, a
municipality should implement. The options range
from costly investment in new infrastructures that
often
source
water
from
neighbouring
jurisdictions, to improved maintenance of existing
infrastructure and water saving practices, to
various economic incentives to reduce water use,
such as metering and fees based on amount used.
The baseline data presented in this bulletin
indicate that, on average, water used for domestic
purposes is higher in rural municipalities than in
urban ones, but that rural households who drink
tap water1 are less likely to treat the water than
urban households. Both the amount used and the
perception of quality perception might be affected
by location characteristics. On the one hand, the
sparse settlement patterns of rural areas offer
more opportunities for recreational uses of water
(pools, gardens, etc), which in turn could explain
the higher amounts used. On the other hand, a
relatively high reliance on private sources of
water (in particular wells) or different socioeconomic characteristics of rural households
could be a key factor in explaining higher
confidence in the quality of water used at home.
Two data sources are used in the analysis: the
Municipal Water and Wastewater Survey
(MWWS) database 2004 of Environment Canada
(Environment Canada, 2007b) and the
Households and the Environment Survey, 2006 of
Statistics Canada (Statistics Canada, 2006)
(Box 1).
This growth in water demand has been paralleled
by growing concern about the quality of domestic
water supply. In the 1990s, several cases of
municipal water contamination received extensive
media coverage and raised public awareness about
the importance of water quality control. As a
result of these cases, some provincial
governments adopted specific legislation that set
minimum standards and testing procedures for
municipal water (Brennan, 2005). However, the
understanding of the relationship between
household perception and household practices
regarding water quality is not well documented.
This analysis focuses on the effect of “rurality” in
determining:
1. per capita water use at the municipal
level; and
2. water quality perception of a
household, as proxied by the water
treatment choice of a household.
Statistics Canada – Catalogue no. 21-006-X
1. “Tap water” is literally water from a tap in the house,
regardless of the source of the water (municipal system,
private well, etc.)
3
Rural and Small Town Canada Analysis Bulletin, Vol. 7, No. 5
Box 1 Data source
This analysis is based on two data sources: the Municipal Water and Wastewater Survey (MWWS),
2004 (Environment Canada, 2007b) and the Households and the Environment Survey, 2006 (Statistics
Canada, 2006).
Municipal Water and Wastewater Survey (MWWS) 2004, produced by Environment Canada, was
based on the previous Municipal Water Use (MUD) survey. Compared to its predecessor survey,
MWWS 2004 introduced several changes in the questionnaire and, for the first time, included returns
from 660 rural municipalities and small incorporated towns with less than 1,000 residents
(Environment Canada, 2007a). As indicated by Environment Canada (2007a), the total usable survey
base for 2004 includes 1,418 municipalities, representing about 28.9 million Canadians (or about 90%
of the estimated Canadian population for that year). However, almost half of these municipalities were
included in the database only after imputation and adjustment of previous MUD data. The response
rates of MWWS 2004 varied substantially by question (Environment Canada, 2007a).
The database used for the computations in this paper retained 1,009 municipalities, for which data on
the variables of interest were available. Appendix Table A.1 shows the count and population distribution
of these municipalities by type of region. Some additional observations with invalid responses on
specific variables were further excluded from the multivariate analysis. The MWWS 2004 variables
were weighted by the population served by the municipal system to obtain summary statistics.
The Households and the Environment Survey (HES) 2006, conducted by Statistics Canada, was
based on the sample frame of the Labour Force Survey and had a sample size of 28,334 households.
Households were categorized into Census Metropolitan Areas (CMAs), Census Agglomerations (CAs),
and non-CMA/CA areas (Box 2). Rural households were defined as any household that is not within a
CMA or CA. All HES 2006 variables used in this analysis were weighted using survey weights, both
for summary statistics and for regressions analysis. Given the complex sampling design of the HES
2006, standard errors and levels of significance for the estimates were calculated using a bootstrap
method of re-sampling (Box 3).
For further details on these data sources see Environment Canada (2007a and 2007b) and Statistics
Canada (2006).
4
Statistics Canada – Catalogue no. 21-006-X
Rural and Small Town Canada Analysis Bulletin, Vol. 7, No. 5
Box 2 Definitions: geography and regional types
This analysis uses the Rural and Small Town (RST) definition for rural areas, which distinguishes
between Census Metropolitan Areas, Census Agglomerations and Rural and Small Town areas; this is
complemented by a classification of rural areas into Metropolitan Influenced Zones (MIZ) (du Plessis
et al., 2002). Incorporated towns and municipalities (Census Subdivisions or CSDs) are the building
blocks for the delineation of CMAs, CAs and the MIZ zones. Hence, we applied this regional typology
to the CSDs in the MWWS 2004 database.
It should be noted that MWWS 2004 extended the survey to rural municipalities with population of
less than 1,000 individuals (the previous MUD included only municipalities with a population of over
1,000). However, in the database used for this analysis, these CSDs represent only about 5% the
sample. In consequence, the statistics for rural should be interpreted primarily as Rural and Small
Town municipalities with 1,000 or more residents and which are not part of a CMA or a CA. Appendix
Table A.1 provides further details on the distribution of municipalities in the database by type of region.
A Census Metropolitan Area (CMA) has an urban core population of 100,000 and over. In this paper,
CMAs have been divided into three groups based on population size: (1) Larger CMA, with
population greater than 1.5 million (these are Toronto, Montreal and Vancouver); (2) Medium CMA,
with population of 0.5 to 1.5 million people; and (3) Smaller CMA, with a population of less than 0.5
million people. A Census Agglomeration (CA) has an urban core population of 10,000 to less than
100,000. Both CMAs and CAs include all neighbouring municipalities where 50% or more of the
workforce commutes to the urban core.
Rural and Small Town (RST) areas are towns or municipalities outside the commuting zone of
CMAs and CAs. RST areas are disaggregated into four Metropolitan Influenced Zones (MIZ) based
on the size of commuting flows of the workforce to any CMA or CA. The strong MIZ category
comprises areas with a commuting flow of 30% or more. The moderate MIZ category comprises areas
with a commuting flow between 5% and less than 30%. The weak MIZ category comprises areas with
a commuting flow of more than 0% and less than 5%. The no MIZ category comprises those areas
where no individuals commute to a CMA/CA; this category is collapsed with the weak MIZ group due
to the small number of observations in the database.
To maintain consistency in the reporting of our results, we also present the results of the Households
and the Environment Survey in terms of CMAs, CAs and MIZ zones.
We recognize that there are census urban residents (in settlements of 1,000 or more) and census rural
residents (in the countryside and in settlements smaller than 1,000) within each CMA, CA and MIZ (du
Plessis et al., 2002). Consequently, our use of the CSD as the geographic unit of analysis will
sometimes include residents of a town with a municipal water provider plus nearby countryside
residents who would obtain their water from a private source.
Statistics Canada – Catalogue no. 21-006-X
5
Rural and Small Town Canada Analysis Bulletin, Vol. 7, No. 5
Box 2 Definitions (continued): water consumption and measurement
The key definitions concerning water use are indicated below. For more details see also Environment
Canada (1999 and 2007b) and Statistics Canada (2006).
Water destination. Three destinations or “sectors” are shown in the MWWS database: domestic (i.e.
all household users); commercial/industrial (i.e. all manufacturing and businesses uses); and other
which includes system losses and unaccounted uses (which, according to Environment Canada (1999),
is believed to be under-reported). The total water use is the sum of these three destinations.
Measure of water used. Average daily total water use per capita (also referred to as total water use
per day, per capita) is calculated by dividing the total daily water flow in each area (for example, CMA
or strong MIZ) by the population served. Average daily domestic water use per capita (also referred
to as domestic water use per day, per capita) is the water flowing through municipal water systems that
is consumed by domestic or household users. These estimates are generated by dividing the domestic
water flow of each municipal system by the population served.
Source of domestic water consumption. Municipal systems are water distribution centers that are
run by municipalities and distribute water to residents who are connected to the system. Private wells
are individual systems (usually paid for by the owner of the property) used by the private owner to
obtain water.
Water metering. The degree of metering refers to the percent of total water flow that is metered for
purposes of monitoring water use (used to determine the water use fee). It is estimated by municipal
managers. Areas classified as “high metering” are municipal systems with more than 90% of serviced
households being metered. Areas classified as “some metering” reported metering for 10% to 90% of
serviced households. Areas classified as “low metering” reported less than 10% of serviced
households being metered.
Water treatment. the HES 2006 considers any activity undertaken by a household to treat their water
supply to be water treatment. A filter or purifier on a household water tap is a device that attaches to
a water tap for the purpose of treating water. A stand alone filter is any device that is not connected to
a water system (municipal system or private well) that has the purpose of treating water. This can range
from an ordinary simple charcoal filter to a more complex osmosis filtration system.
6
Statistics Canada – Catalogue no. 21-006-X
Rural and Small Town Canada Analysis Bulletin, Vol. 7, No. 5
Water uses, flows and source
There are substantial differences among types of
regions, in terms of municipal water uses, the
quantity used, and the source of the water that is
used by households. Generally, the share of
domestic water use relative to the total use and the
amount used per capita are both higher in rural
areas. Although water from a municipal system is
the main source in each type of region, about one
third of rural households rely on private sources of
water (i.e. wells) for domestic use.
The geographic classification used in this analysis
compares urban centers (CMA/CA) and rural and
small town (RST) areas (Box 2). In some
tabulations, a distinction is also made between
towns and municipalities classified within larger
agglomerations, namely Toronto, Montreal and
Vancouver, and towns and municipalities
classified
within
medium
and
smaller
agglomerations (Box 2).
It should be recalled, however, that the database
generated from MWWS 2004 for the purpose of
this analysis includes only 5% of municipalities
with less than 1,000 residents (Box 2). Moreover,
relatively few non-farm households in the
countryside (and even fewer farms) are connected
to a municipal water provider. These households
rely primarily on private wells and, as a result, the
MWWS database does not provide statistics on
water used by them.
The MWWS 2004 classifies municipal water use
by three main destinations: domestic or household
use, industrial and commercial use, and water loss
or leakage from the municipal water system (Box
2). Overall the patterns of use in larger urban
centers and rural and small towns are similar.
However, the share of water devoted to domestic
use is generally higher in rural and small town
areas (Figure 1). The share of water used for
commercial and industrial uses is the lowest in
strong MIZ and highest in the smaller CMAs. The
share of other uses, which includes water loss
Statistics Canada – Catalogue no. 21-006-X
from the system, is slightly lower in rural areas
(less than 10%) compared to larger urban centers
(generally over 10%).
When water use is disaggregated by type of area
with the MWWS 2004 dataset, the results indicate
that per capita water use is higher in
municipalities that are more rural (Figure 2). This
pattern is true for total water used and water used
for domestic purposes. Total water use per capita
is lower (below 500 litres) in larger urban centers
(larger CMA and medium CMA); it grows to
about 600 litres in smaller CMA, CA and strong
MIZ; and it is higher in zones that are more rural
– over 800 litres per day per person in moderate
and weak/no MIZ.
Domestic water use per capita shows a pattern
similar to total water use per capita. The daily use
is estimated at about 300 litres per capita for
residents of all types of urban areas (Figure 3).
However, medium CMAs are appreciably below
that average (about 250 litres).
Among the ten provinces, in 2004 Manitoba
showed the lowest domestic daily use of water at
219 litres per capita, while Newfoundland and
Labrador showed the highest domestic daily use
at 501 litres per capita (Environment Canada,
2007a). Provincial patterns across the urban-torural gradient tended to be similar to the national
pattern (data not shown), but in some provinces
the number of observations by regional type is not
large enough to provide for a robust sample at the
sub-provincial level.
The sources of water for domestic use are
typically municipal supply, private wells, and
surface water. Although the municipal water
supply is prevalent in all regional types, there are
still large differences between the residents of
rural and urban municipalities. In rural areas,
about one-third of households relied on water
from a private well in 2006 (Figure 4). Rural
households cannot always access a municipal
water system and need to drill their own private
7
Rural and Small Town Canada Analysis Bulletin, Vol. 7, No. 5
well or seek alternative water sources.
Comparatively, just over 14% of households
within CAs obtain their water from a private well,
and fewer than 4% of households in CMAs use a
private well. However, two-thirds of rural
households obtain their water from a municipal
system, making municipal systems the main
source of drinking water across Canada.
Figure 1 Municipal water systems in rural and small town areas report a higher share of water
being used for domestic use
percent distribution of municipal water flow by type of destination, Canada, 2004
100
80
60
40
20
0
Larger census
metropolitan
area
Medium census
metropolitan
area
Smaller census
metropolitan
area
Larger urban centres
Domestic
Census
agglomeration
Strong
Moderate
Weak/No
metropolitan
metropolitan
metropolitan
influenced zone influenced zone influenced zone
Rural and small town areas
Commercial/ industrial
Other/loss
Note: Figures are for a sample of 1,009 municipalities used in this analysis. Shares are weighted by total flow.
Source: Environment Canada, Municipal Water and Wastewater Survey (MWWS), 2004.
8
Statistics Canada – Catalogue no. 21-006-X
Rural and Small Town Canada Analysis Bulletin, Vol. 7, No. 5
Figure 2 Municipal water use for all purposes, on a per capita basis, is highest in weak/no
metropolitan influenced zone
municipal water flow for all uses (litres per capita per day), Canada, 2004
1000
900
800
700
600
500
400
300
200
100
0
Larger census
metropolitan
area
Medium census
metropolitan
area
Smaller census
metropolitan
area
Larger urban centres
Census
agglomeration
Strong
Moderate
Weak/No
metropolitan
metropolitan
metropolitan
influenced zone influenced zone influenced zone
Rural and small town areas
Note: Figures are for a sample of 1,009 municipalities used in this analysis. Averages are weighted by municipal population served. The
sample average is 564 litres perr capita per day, compared to the 609 litres per capita per day for the entire Municipal Water and
Wastewater Survey database (Environment Canada, 2007a).
Source: Environment Canada, Municipal Water and Wastewater Survey (MWWS), 2004.
Statistics Canada – Catalogue no. 21-006-X
9
Rural and Small Town Canada Analysis Bulletin, Vol. 7, No. 5
Figure 3
Domestic water use of municipal water systems is highest in moderate metropolitan
influenced zone
municipal water flow for domestic use (litres per capita per day), Canada, 2004
600
500
400
300
200
100
0
Larger census
metropolitan
area
Medium census
metropolitan
area
Smaller census
metropolitan
area
Larger urban centres
Census
agglomeration
Strong
Moderate
Weak/no
metropolitan
metropolitan
metropolitan
influenced zone influenced zone influenced zone
Rural and small town areas
Note: Figures are for a sample of 1,009 municipalities used in this analysis. Averages are weighted by municipal
population served. The sample average is 320 litres per capita per day, compared to the 329 litres per capita per day
for the entire Municipal Water and Wastewater Survey database (Environment Canada, 2007a).
Source: Environment Canada, Municipal Water and Wastewater Survey (MWWS), 2004.
10
Statistics Canada – Catalogue no. 21-006-X
Rural and Small Town Canada Analysis Bulletin, Vol. 7, No. 5
Figure 4 One-third of rural households obtain their water from a private well
percent distribution of households by source of household water, Canada, 2006
100
80
60
40
20
0
Census metropolitan area
Municipal supply
Census agglomeration
Private well
Rural and small town
Surface source
Note: “Other sources” are not included as their incidence is marginal.
Source: Statistics Canada, Households and the Environment Survey (HES), 2006.
Statistics Canada – Catalogue no. 21-006-X
11
Rural and Small Town Canada Analysis Bulletin, Vol. 7, No. 5
Explaining differentials in water use
Various hypotheses could be formulated about the
determinants of the amount of water used for
domestic purposes, as presented in the previous
section (for a review see Arbués et al., 2003 and
Dalhuisen et al., 2003). The existing literature has
pointed in particular to two sets of explanatory
factors: locational characteristics and economic
incentives. Rural municipalities are characterized
by low population density which may offer more
opportunities for domestic water use (Troy and
Randolph, 2006). Rural properties are commonly
larger than urban properties and lawn and garden
watering that nourish greenery is a typical waterconsuming amenity in this environment. Hence, it
would be expected that increasing rurality would
be associated with higher use of water.
In terms of economic incentives, evidence on the
consumption response to water pricing is
presented by Dandy et al. (1997), Dalhuisen et al.
(2003) and Reynaud et al. (2005). In the Canadian
context, it has been suggested that users pay too
little for their water. As argued by Environment
Canada (2001), one of the key factors determining
high water use is the absence of appropriate price
signals. Environment Canada (2007a) indicates
that, in 2004, approximately 63% of Canadians
served by municipal water systems were metered.
(It is 70% for the sample used in this analysis.)
For the sample of municipalities used in this
analysis, the share of the population in rural areas
that is metered is less than 60% (Figure 5). This
may be an important factor in the higher average
water use in rural areas and this finding is
consistent with the existing literature.
Also, water metering policies differ across
provinces. Specifically, less than 8% of municipal
water users in Newfoundland and Labrador,
Prince Edward Island and the Yukon are metered.
Also, 16% of municipal water users in Quebec are
metered and 30% of municipal water users in
British Columbia are metered (Environment
Canada, 2007a). Although there is a wide
12
variation in the rate of water metering among
municipalities in Quebec, Ontario and British
Columbia, Canada’s three largest cities broadly
reflect that of their respective provinces.
Households in the Montreal CMA have
essentially the same incidence of water metering
as the other residents throughout Quebec (about
16%). In the Toronto CMA, about 98% of
households have water meters while the figure is
92% for Ontario as a whole. Meanwhile, about
15% of households in the Vancouver CMA have a
water meter compared to 30% throughout British
Columbia, where other large municipalities have
high metering rates2.
To assess the effect of locational and economic
factors in determining the per capita domestic use
of water at the municipal level, a regression
model is applied (Box 3). Location factors include
the municipal population density in 2001
(inhabitants per square kilometre), and the type of
region (Box 2). The effect of economic incentives
is captured by three dummy variables indicating
the extent of municipal water metering: “high
metering” indicates that more than 90% of the
population served is subject to metering, “low
metering” indicates that less than 10% of the
population relying on municipal water supply is
subject to metering and “some metering” covers
the range between the high and low cases. The
low metering category was used as the reference
group and therefore dropped from the model.
Appendix Table A.3 shows descriptive statistics
and Appendix Table A.4 shows the results of the
regression models.
The dataset used in this analysis includes a total
of 963 municipalities for which the indicators
used in the model were available. Two alternative
specifications of the model were used. The first
2. Note that each CMA and CA typically has many
municipalities included within their boundaries (Box 2)
and many of these municipalities would have their own
water service. In 2001, there were 50 census
subdivisions (i.e. incorporated towns and municipalities)
within the Montreal CMA, 17 within the Toronto CMA
and 18 within the Vancouver CMA.
Statistics Canada – Catalogue no. 21-006-X
Rural and Small Town Canada Analysis Bulletin, Vol. 7, No. 5
included only population density as a location
variable while the second included the regional
indicator variables as well. Both specifications
were also estimated with a sub-sample of 865
observations which excluded outliers that had
extreme high and low values of water use. The
estimated relationship that excluded the highest
and lowest values tested whether the results were
sensitive to a few municipalities whose water use
was either considerably higher or considerably
lower than the average.
The regression results suggest that economic
incentives, as measured by the degree of
municipal metering, are generally more important
than locational factors in affecting average
domestic use of water3. Once the location
characteristics of the municipality are controlled
for, municipalities with high degree of water
metering are associated with an average daily
water use of about 170 litres less than a similar
municipality with low metering (for the subsample with the high and low outliers removed),
or about 200 litres less when the full sample is
used. Shifting from low metering to some
metering is associated with a lower average
domestic water use of about 70 litres per person
per day or approximately 20% of 2004 average
water use.
As mentioned above, the existing literature and
economic theory suggest that the higher the price
of the water the less a household will use of it. It
is therefore not surprising that the presence of a
water meter tends to reduce domestic water use.
Population density also has a statistically
significant association with water use levels, as
expected. For each additional 100 inhabitants per
square kilometre, water use is lower by about 50
to 90 litres (depending on the specification). In
contrast, locational characteristics of the
municipality, as captured by the type of area,
appear to have a low association with water use.
A statistically significant effect is found only for
weak/no MIZ (in the full model and at the 10%
statistical significance level). The daily per capita
water use in this type of area is higher than in
CMAs (which is the reference group). The
magnitude of the increase is approximately 60
litres per capita per day.
Again, given that the existing literature shows that
there are greater opportunities for domestic water
use on rural properties, these results would be
expected.
In summarizing to this point, the results indicate
that average daily domestic water flow is lower in
municipalities that meter their water flow – that
is, economic incentives are clearly associated with
level of water use. The population density of the
settlement is also associated with the level of
water use. In contrast, the type of area has a more
limited association with water use. However,
when significant, the results are still consistent
with the expectation of higher water use in rural
areas.
3. Reynaud et al. (2005) performed an analysis of 899
municipalities which pooled data from the 1993, 1995
and 1998 versions of the MUD database. They
determined that the estimated change in the quantity of
water consumed in response to a different price is
somewhat over-estimated if one does not first take into
account that the decision to introduce a water pricing
structure is influenced by the characteristics of the
municipality. Thus, our results may overstate
(somewhat) the estimated lower level of water
consumption due to the presence of water metering.
Statistics Canada – Catalogue no. 21-006-X
13
Rural and Small Town Canada Analysis Bulletin, Vol. 7, No. 5
Figure 5 In rural and small town areas, less than 60% of the population serviced by a municipal
water system has water meters
population metered as a percent of population served by the municipal water system, Canada, 2004
100
80
60
40
20
0
Larger census
metropolitan
area
Medium census
metropolitan
area
Smaller census
metropolitan
area
Census
agglomeration
Larger urban centres
Strong
Moderate
Weak/No
metropolitan
metropolitan
metropolitan
influenced zone influenced zone influenced zone
Rural and small town areas
Note: Figures are for a sample of 1,009 municipalities used in this analysis. Averages are weighted by municipal
population served. The average percentage of population metered for this sample is 71%, compared to the entire
Municipal Water and Wastewater Survey database where 63% of residential clients and 83% of business clients were metered
(Environment Canada, 2007a).
Source: Environment Canada, Municipal Water and Wastewater Survey (MWWS), 2004.
Water treatment as a possible indicator
of water quality perceptions
There is evidence that suggests that water quality
is a growing concern for many Canadians
(Adamowicz et al., 2007). Yet, there is limited
understanding of the relationship between water
quality concerns and water consumption decisions
taken by Canadian households. In this part of the
analysis we look at the choice of a household to
14
treat its tap water. Although this does not address
the issue of water quality perception directly,
consumption choice remains a plausible, albeit
indirect, indicator of water quality perceptions.
The results of the Households and the
Environment Survey (HES) 2006 show that
between 23% and 31% of households primarily
drink bottled water, depending on location
(Appendix Table A.2). For these households, the
HES 2006 does not collect information on water
Statistics Canada – Catalogue no. 21-006-X
Rural and Small Town Canada Analysis Bulletin, Vol. 7, No. 5
treatment and therefore these households are not
included in the analysis (see Box 2 for definitions
of water treatment). When we excluded
households that primarily drink bottled water,
only about 40% of rural households use some
form of water treatment, compared to over 50% in
urban areas (Figure 6). This figure includes both
rural households that are connected to municipal
water systems as well as those which rely on
private wells. Rural households using private
wells are not notably different from rural
households that are using municipal systems when
it comes to the decision of whether or not to treat
their drinking water.
reference group), while households in rural areas
are only 0.7 times as likely to treat their drinking
water compared to the reference group.
In this analysis we focus on the factors associated
with water treatment choices by households that
drink primarily tap water or that drink both tap
water and bottled water (that is, households that
drink primarily bottled water are excluded from
the analysis). The results are based on a logit
model (Box 3). Using four sets of variables, this
model shows the association of each variable with
the probability that households are treating their
drinking water. The four sets of explanatory
variables are: (1) demographic variables, which
included family size and age cohorts of the
household
members;
(2)
socio-economic
variables, which include educational attainment
and income level; (3) type of water source; and
(4) locational characteristics of the households,
defined as rural and major type of urban
agglomerations (Box 2). Descriptive statistics for
these explanatory variables are presented in
Appendix Table A.5. The survey sample includes
15,504 households representing about 6.8 million
Canadian households.
Demographic factors have some effect but their
patterns appear more difficult to interpret. The
presence of an individual age 16 to 24 or the
presence of an individual 55 years and over
increases the likelihood of water treatment, while
the number of individuals in other age groups has
no significant effect. Finally, the source of water
(municipal or private) does not have a significant
effect on treatment choices of the households. In
other words, the household that relies on a private
source of water is not different in its water
treatment decisions from those that rely on
municipal water supply systems.
The detailed results of the logit model, including
bootstrapped standard errors are reported in
Appendix Table A.6. These results indicate that
location plays a large role in the household
decision to treat drinking water. The odds of
observing water treatment are higher by a factor
of 1.7 for households located in Toronto
(compared to the medium and smaller CMA
Statistics Canada – Catalogue no. 21-006-X
Socio-economic characteristics of the household
also have a significant role, although this is only
for specific categories. Households with any
member with a high level of educational
attainment (university degree level) and
households with a higher income level
(specifically between $80,000 and $99,999) are
more likely to treat their water (in both cases the
odds of water treatment are higher by a factor of
1.3).
The results of the model allow the prediction of
the probability of observing water treatment for
typical household profiles. For the purpose of this
analysis and for benchmarking purposes, the
typical household is defined as a household with
three members, two aged 45 to 54 and one aged
16 to 24, high school graduation (but no postsecondary) as the highest level of education
attained by any household member, average
household income between $50,000 and $59,999,
using municipal water and drinking only tap water
Figure 7 shows the predicted probability of water
treatment for this typical Canadian household in
different locations in Canada (See Appendix
Table A.7 for details).
The geographic location of this typical household
had a large impact on the likelihood of treating
15
Rural and Small Town Canada Analysis Bulletin, Vol. 7, No. 5
water. A household with this typical profile that is
located in the large CMAs of Vancouver and
Toronto has about a 55% chance of treating water
(Figure 7). Montreal is a notable exception to the
urban-rural gradient – the chance of treating water
was below 30%, following the overall Quebec
pattern of a lower likelihood of treating tap water
(Statistics Canada, 2007). Note the strong urbanto-rural gradient of a lower probability of water
treatment as one compares CMAs to CAs and as
one compares CAs to rural areas.
Figure 6 In rural areas, 40% of households treat their drinking water
percent of households who treat their drinking water
(excluding households who primarily drink bottled water), Canada, 2006
60
50
40
30
20
10
0
Census metropolitan area
Census agglomeration
Rural and small town
Source: Statistics Canada, Households and the Environment Survey (HES), 2006.
The predicted probabilities of this typical
household are used as a benchmark for other types
of household profiles. In particular, four additional
household profiles are identified. These profiles
emphasize household differences in the variables
of the logit model that were statistically
significant. Details on the specification of each
16
profile and the resulting predicted probability by
area of residence are reported in Appendix Table
A.7.
The pattern in predicted probabilities for these four
household profiles follows that observed for the
base case – there is a clear urban-to-rural gradient
Statistics Canada – Catalogue no. 21-006-X
Rural and Small Town Canada Analysis Bulletin, Vol. 7, No. 5
for each profile. (Montreal, again, is an exception,
following the lower likelihood of households in
Quebec to treat their tap water.) Other household
characteristics being the same, rural households
were less likely to treat their water than urban
households. Compared to the typical household,
the single household (Case 2) has only one
household member, has some post-secondary
education, earns between $30,000 and $39,999 per
year, and obtains drinking water from a private
source. When this household is located in Toronto
the probability of water treatment is 54%, whereas
the same type of household located in rural areas
has a 33% probability of treating water (Appendix
Table A.7). The elderly couple household (Case 3)
emphasizes the age variable within the household
with an above average household income. Again,
the urban areas of Toronto and Vancouver have a
greater likelihood of treating their water than rural
areas. This trend is also present in the other two
cases: the mixed age, high education household
(Case 4), which captures a higher household
education attainment; and the young professional
household (Case 5), which emphasizes the effect
of high income.
The differences in predicted probability of water
treatment between the typical household and other
household profiles are further detailed in Figure 8.
This chart is evidence of the role of other
Statistics Canada – Catalogue no. 21-006-X
variables, in addition to geographic location, in
determining the probability of water treatment.
For instance, compared to the typical household,
the single household has a 1.8 to 2.3 percentage
point lower probability of treating water within
each area. For each case, the difference in
estimated probabilities (compared to the typical
household) is consistent across the geographic
areas. A consistent difference compared with the
typical case means that each case has an urban-torural gradient that is just as strong as the gradient
in the typical case (Figure 7).
A household with the mixed family, higher
education profile (Case 4) has a 13 to 14
percentage point higher probability of treating
drinking water. Hence, differences in demographic
and socio-economic characteristics of the
households have a large and significant effect on
the probability of water treatment.
Importantly, the difference in the share of
households treating drinking water is flat across
the urban-to-rural gradient (Figure 8). That is, for
any given case, the percent of households treating
drinking water shows the same strong urban-torural gradient as in Case 1 (the gradient is
essentially the same because the difference with
Case 1 is essentially the same).
17
Rural and Small Town Canada Analysis Bulletin, Vol. 7, No. 5
Figure 7 For a typical household, there is a strong urban to rural gradient in the probability of
households treating their drinking water
probability of a typical household (Case 1 in Appendix Table A7) treating its drinking water (excluding households
who primarily drink bottled water), Canada, 2006
60
50
40
30
20
10
0
Montreal
Toronto
Vancouver
All other census
metropolitan areas
Census
agglomerations
Rural and small
town
Note: See Box 3 for methods used to derive figures. See Appendix Table A.7 for a description of each case.
Source: Estimated results based on Statistics Canada, Households and the Environment Survey (HES), 2006.
18
Statistics Canada – Catalogue no. 21-006-X
Rural and Small Town Canada Analysis Bulletin, Vol. 7, No. 5
Figure 8 Each selected household profile has the same strong urban to rural gradient as the typical
household
percentage point difference in the probability of treating drinking water - each case is compared to the
typical household (Case 1)
20
15
10
5
0
Montreal
Vancouver
Census agglomerations
-5
Toronto
All other census metropolitain areas
Rural
-10
Case 2: Single
Case 3: Elderly couple
Case 4: Mixed age, high
education
Case 5: Young professional
Household profiles
Note: See Box 3 for methods used to derive these figures. See Appendix Table A.7 for a description of each case.
Source: Estimated results based on Statistics Canada, Households and the Environment Survey (HES), 2006.
Statistics Canada – Catalogue no. 21-006-X
19
Rural and Small Town Canada Analysis Bulletin, Vol. 7, No. 5
Conclusions
This bulletin focuses on the effect of “rurality” in
determining the quantity of water used, at the
municipal level, and on water quality perception
of households, as measured by a household’s
water treatment choices. The baseline data
presented in this bulletin indicates that, on
average, water used for domestic purposes is
higher in rural municipalities than in urban ones,
but that rural households who drink tap water are
less likely to treat their water than urban
households. Urban households are also more
reliant on municipal water systems than rural
households. Though municipal water systems are
the main source of water for the majority of rural
households, private wells remain an important
source of water as well.
The results of this analysis suggest that economic
incentives are more relevant than location
characteristics in determining average water use.
Areas with higher shares of water metering use
less water than areas with lower shares of water
metering. The effect of population density and
regional type are less clear, although they are to
some extent consistent with the expectation that
areas that are more rural would have higher water
use compared to urban residents because of the
tendency to have larger gardens that need summer
watering.
In contrast, locational characteristics are
significant determinants of the choice of treating
tap water for domestic consumption while the
source of water (municipal or private) is not. The
location effect on treatment choices remains
strong even after controlling for socio-economic
characteristics of the household. However, some
socio-economic characteristics also have a
significant impact on the likelihood of a
household treating its water.
Rural households appear more confident in the
quality of their water supply, compared to their
urban counterparts. Almost 60% of rural
households do not treat their water, whereas fewer
than 50% of CMA households do not treat their
water. Finally, the type of water consumed
(primarily tap or a combination of tap and bottle
water) does not appear to be a factor in the
treatment decision.
There are several implications for municipal water
strategies. First, the results suggest that economic
incentives might play a significant role for
municipal water management. Even though the
introduction of water metering systems cannot be
viewed as a substitute for infrastructure policies,
this could remain an important measure in
containing the quantity of water used.
Second, assuming that the choice of water
treatment reflects quality perception, municipal
water sources are not perceived to have higher
quality as compared to private sources. Household
characteristics, such as higher education and
income level, explain in part the choice of treating
the water, but there appears to be a location
specific factor that should be further investigated.
With the notable exception of Montreal, urban
households are more sensitive to water quality
issues, as reflected by the decision to treat tap
water for domestic consumption.
David Hardie is a co-op student at the University of Waterloo and
Alessandro Alasia is an analyst in the Research and Rural Data Section,
Agriculture Division.
20
Statistics Canada – Catalogue no. 21-006-X
Rural and Small Town Canada Analysis Bulletin, Vol. 7, No. 5
References
Adamowicz, V., D. Dupont, A. Krupnick and P.
Payment. (2007) Determinants of Water
Consumption Choices: an Ordered Probit
Approach. Paper presented to the 42nd
Central Canadian Symposium on Water
Quality Research, Environment Canada.
Arbués, Fernando, María Ángeles García-Valiñas,
and Roberto Martínez-Espiñeira. (2003)
Estimation of residential water demand: a
state-of-the-art review, Journal of SocioEconomics 32:81-102.
Brennan, Richard. (2005) “Safe water legislation
introduced.”
The
Toronto
Star.
December 6.
Dalhuisen, Jasper M., Raymond J.G.M. Florax,
Henri L.F. de Groot, and Peter Nijkamp.
(2003) Price and Income Elasticities of
Residential Water Demand: A Metaanalysis ; Land Economics, 79(2) 292-308
Environment Canada. (2007a) Municipal Water
Use, 2004 Statistics. Environment Canada
Catalogue no. En11-2/2004E-PDF.
Environment Canada. (2007b) Municipal Water
and Wastewater Survey – MWWS.
Variable
Description
Document.
Environmental
Stewardship
Branch,
Environment Canada.
Long, Scott J., and Jeremy Freese. (2001)
Regression Models for Categorical
Dependent Variables Using Stata. College
Station, U.S.A. Stata Press.
Munro, Margaret. (2004) “Alberta faces water
shortage: An environmental scientist says
the province must put the brakes on its
energy boom and population growth.” The
Vancouver Sun. Canwest News Service.
April 4.
Dandy, Graeme, Tin Nguyen, and Carolyn
Davies. (1997) Estimating water demand
in the presence of free allowances. Land
Economics 73(1), p. 125-139.
Reynaud, A., S. Renzetti, and M. Villeneuve.
(2005) Residential water demand with
endogenous pricing: The Canadian Case.
Water Resources Research Vol. 41,
W11409, doi:10.1029/2005WR004195.
du Plessis, Valerie, Roland Beshiri, Ray D.
Bollman, and Heather Clemenson. (2002)
Definitions of Rural. Statistics Canada.
Catalogue no. 21-601-MIE. Ottawa.
Agriculture and Rural Working Paper
Series, no. 61.
Statistics Canada. (2006) Households and the
Environment Survey (HES) Detailed
Information for 2006. Statistics Canada
Record no. 3881. Ottawa. Modified on
November 10, 2006. Ottawa.
Environment Canada. (1999) Municipal Water Use
Database. Environment Canada.
Environment Canada. (2001) Municipal Water
Pricing, 1991-1999. Environment Canada.
Environment Canada. (2005) Municipal Water
Use, 2001 Statistics. Environment Canada
Catalogue no. En11-2/2001E-HTML.
Updated February 21, 2005.
Environment Canada. (2006) Data Description –
2001 Municipal (Water) Use Database
(MUD). Updated October 3, 2006.
Statistics Canada – Catalogue no. 21-006-X
Statistics Canada. (2007) Households and the
Environment: 2006 (Ottawa: Statistics
Canada, Catalogue no. 11-526).
The Walter and Duncan Gordon Foundation.
(2004) Controlling Our Thirst: Managing
Water Demands and Allocations in
Canada. Toronto. Canadian Science
Writers’ Association Conference, June
2004.
The Whig-Standard. (2006) “Water system gets
attention; Plan could lead to infrastructure
21
Rural and Small Town Canada Analysis Bulletin, Vol. 7, No. 5
upgrades, changes.” The Kingston WhigStandard. Kingston, ON. November 28.
Troy, Patrick, and Bill Randolph (2006). Water
Consumption and the Built Environment:
A Social and Behavioural Analysis.
Research Paper No. 5. City Futures
Research Centre. University of New South
Wales.
Another Statistics Canada innovation…
Readers may also be interested in: EnviroStats (Catalogue no. 16-002-X)
EnviroStats is Statistics Canada’s quarterly bulletin of environmental and sustainable development statistics.
EnviroStats provides regular statistical analysis of environmental topics written for a broad audience. At the
core of each issue is a feature article. Shorter articles highlight new statistical developments or introduce
new concepts. “Updates” cover recent and upcoming events such as releases of new statistical products or
overviews of surveys under way. An extensive data table ensures that readers have the most recent
statistics available. Each issue will also feature a map illustrating and analyzing a current topic.
Statistics Canada http://www.statcan.gc.ca/bsolc/english/bsolc?catno=16-002-X.
22
Statistics Canada – Catalogue no. 21-006-X
Rural and Small Town Canada Analysis Bulletin, Vol. 7, No. 5
Appendix Table A.1 Population and communities included in the Municipal Water and Wastewater
Survey
Population
Communities
Municipal
Water and
Wastewater
Survey
2004
analysis
percent
71.2
69.9
76.8
Municipal
Water and
Wastewater
Survey
2004 analysis
Canada
individuals
23,839,086
16,971,853
19,296,926
13,483,090
4,542,160
3,488,763
Type of region
Larger urban centres
Census metropolitan area
Census agglomeration
Rural and small town
areas
Strongly influenced
Moderately influenced
Weakly influenced
Not influenced/Territories
All regions
Municipal
Water and
Wastewater
Survey
2004 analysis
Canada
communities
990
356
468
188
522
168
Municipal
Water and
Wastewater
Survey
2004 analysis
percent
36.0
40.2
32.2
6,168,008
1,524,579
2,285,538
1,969,211
388,680
2,624,730
625,833
924,103
1,009,939
64,855
42.6
41.0
40.4
51.3
16.7
4,580
566
1388
1014
1612
653
96
238
272
47
14.3
17.0
17.1
26.8
2.90
30,007,094
19,596,583
65.3
5,570
1,009
18.1
Note: The columns “Municipal Water and Wastewater Survey 2004 analysis” report the population and community figures for the municipalities of the
Municipal Water and Wastewater Survey 2004 included in our analysis; that is, for the municipalities for which data were available for the variables of
interest. All population figures are for the year 2001.
Source: Authors’ elaboration based on Environment Canada, Municipal Water and Wastewater Survey (MWWS), 2004 and Statistics Canada, Census of
Population, 2001.
Appendix Table A.2 Primary type of drinking water consumed by households, Canada, 2006
Census metropolitan area
Census
Vancouver
Toronto
Montreal
Other agglomeration
percent distribution of households
Type of drinking water
Primarily tap water
Combination drinking water
Primarily bottled water
1
Missing2
Rural
66.3
53.4
58.7
58.0
55.3
57.1
10.2
23.0
18.8
26.8
10.2
30.3
12.4
28.4
12.2
31.0
10.6
29.4
0.3
0.1
0.2
0.1
0.2
0.1
1. “Combination drinking water” refers to households which consume tap and bottled water, including “other” responses.
2. “Missing” includes Refused or Don't Know responses.
Note: All figures are weighted figures. The weighted sample, used for this computation, corresponds to 12,568,539 households. Households which responded
“Primarily bottled water” were not included in the logistic regression analysis. This is because with the Households and the Environment Survey 2006, if a
household primarily drank bottled water, the questionnaire did not request whether the tap water was treated.
Source: Statistics Canada, Households and the Environment Survey (HES), 2006.
Statistics Canada – Catalogue no. 21-006-X
23
Rural and Small Town Canada Analysis Bulletin, Vol. 7, No. 5
Appendix Table A.3 Descriptive statistics: Factors associated with domestic water use from
municipal water systems, Canada, 2004
Variable (mean)
Dependent
Average daily domestic water use in the municipality,
unweighted (litres per capita)
Independent
Metering indicators
High metering (dichotomous)
Some metering (dichotomous)
Low metering (dichotomous)
Location indicators
Population density (inhabitants per km2)
Census metropolitan area (dichotomous)
Census agglomeration (dichotomous)
Strong metropolitan influenced zone (dichotomous)
Moderate metropolitan influenced zone (dichotomous)
Weak/no metropolitan influenced zone (dichotomous)
Number of observations (municipalities)
Mean
Standard deviation
468.30
310.90
0.49
0.07
0.44
0.50
0.26
0.50
339.00
0.19
0.17
0.10
0.24
0.32
482.00
0.39
0.37
0.29
0.42
0.47
1,009
…
Note: Figures are computed from a sample of 1,009 municipalities of the Municipal Water and Wastewater Survey 2004 that reported
information on water metering, total yearly water outflows, and destination flows. Municipalities with water systems outside Whitehorse
and Yellowknife in the Territories are classified as weak/no metropolitan influenced zone municipalities. See Box 2 for variable definitions. All
variables, except average daily domestic use in the municipality and population density, are in dichotomous form, meaning that the variable takes the
value of 1 if the attribute is observed and 0 otherwise. Regional type variables take the value of 1 if the municipality is located in that regional type
(see Box 2 for variable descriptions and definition of types of area).
Source: Environment Canada, Municipal Water and Wastewater Survey (MWWS), 2004.
24
Statistics Canada – Catalogue no. 21-006-X
Rural and Small Town Canada Analysis Bulletin, Vol. 7, No. 5
Appendix Table A.4 Regression results: Variables in regression of factors associated with domestic
water use1 across municipal water systems
Model 1
Variable
Constant
Metering indicators
Some metering
High metering
Location indicators
Population density
Census
agglomeration
Strong metropolitan
influenced zone
Moderate
metropolitan
influenced zone
Weak/no
metropolitan
influenced zone
R2
Number of observations
Full sample
β
t-stat
Omitted outliers
β
t-stat
Full sample
β
Model 2
Omitted outliers
β
t-stat
t-stat
590.07
38.54
535.42
49.73
623.25
18.86
520.16
27.82
-76.03
-200.70
-1.63
-10.48
-106.51
-168.67
-3.57
-12.93
-75.84
-199.25
-1.64
-10.23
-105.47
-167.68
-3.47
-12.61
-0.08
-5.83
-0.05
-5.80
-0.09
-5.51
-0.05
-4.68
…
…
…
…
-20.32
-0.59
16.29
0.78
…
…
…
…
-35.50
-0.86
21.05
0.75
…
…
…
…
-20.14
-0.57
18.06
0.85
…
…
…
…
-57.18
-1.78
13.03
0.69
0.13
963
…
…
0.19
865
…
…
0.13
963
…
…
0.19
865
…
…
1. Litres per capita per day.
Note: The full sample estimation includes 963 municipalities reporting information on water metering, total yearly water outflows and destination flows. The
omitted outliers sample excludes the highest 5% and the lowest 5% of municipalities in terms of domestic water use per capita. Municipalities with
water systems outside Whitehorse and Yellowknife in the Territories are included with weak/no metropolitan influenced zone municipalities. See Box 2
for variable definitions.
Source: Estimation based on Environment Canada, Municipal Water and Wastewater Survey (MWWS), 2004.
Statistics Canada – Catalogue no. 21-006-X
25
Rural and Small Town Canada Analysis Bulletin, Vol. 7, No. 5
Appendix Table A.5 Descriptive statistics: Variables in logistic regression of factors associated with
households which treat drinking water1
Variable name
Dependent variables
Treatment indicator (=1 if household treats drinking water)
Independent variables
Family size
Age of household members (=1 if there is one or more household
members present in the age class)
Aged 0-15
Aged 16-24
Aged 25-44
Aged 45-54
Aged 55+
Education
Some secondary (reference group)
High school graduate (no post-secondary)
Some post-secondary
University degree
Income bracket ($)
0 to 19,999 (reference group)
20,000 - 29,999
30,000 - 39,999
40,000 - 49,999
50,000 - 59,999
60,000 - 79,999
80,000 - 99,999
100,000 and over
Type of water source
Municipal (reference group)
Private
Type of household drinking water
Primarily tap water (reference group)
Combination
Location
Medium and smaller census metropolitan areas (reference group)
Census agglomeration
Rural
Vancouver
Toronto
Montreal
Weighted observations
Sample observations
Mean
Standard deviation
0.473
0.500
2.380
1.378
0.278
0.200
0.479
0.294
0.378
0.448
0.400
0.500
0.456
0.485
0.112
0.123
0.397
0.363
0.315
0.334
0.489
0.481
0.138
0.112
0.124
0.096
0.090
0.155
0.010
0.184
0.345
0.316
0.330
0.295
0.287
0.362
0.300
0.387
0.868
0.132
0.338
0.338
0.811
0.189
0.391
0.391
0.317
0.098
0.237
0.070
0.156
0.120
6,779,154
15,504
0.465
0.297
0.425
0.256
0.363
0.325
…
…
1. Excludes households who primarily drink bottled water.
Note: The descriptive statistics are computed using survey weights. Invalid responses are not included. All variables, except family size, are in dichotomous
form, meaning that the variable takes a value of 1 if the attribute is observed and 0 otherwise. “Reference group” indicates the category that is omitted in the
logistic regression and thus the estimated odds ratios are to be compared to the status of the reference group. Family size is the number of family members
at the time of the survey. The dummy variables for age of household members indicate the presence of one or more household members in that age class –
thus, the variables are not a set of mutually exclusive dummies and no reference group is indicated.
Source: Statistics Canada, Households and the Environment Survey (HES), 2006.
26
Statistics Canada – Catalogue no. 21-006-X
Rural and Small Town Canada Analysis Bulletin, Vol. 7, No. 5
Appendix Table A.6 Logistic regression results: Variables in logistic regression of factors associated
with households which treat drinking water1
Variable
Constant
Family size
Age of household members
Aged 0-15
Aged 16-24
Aged 25-44
Aged 45-54
Aged 55+
Education
High school graduate (no post-secondary)
Some post-secondary
University degree
Income bracket ($)
20,000 - 29,999
30,000 - 39,999
40,000 - 49,999
50,000 - 59,999
60,000 - 79,999
80,000 - 99,999
100,000 and over
Type of water source
Private
Type of drinking water
Combination
Location
Census agglomeration
Rural
Vancouver
Toronto
Montreal
Coefficient
Bootstrapped
standard
error
P-value
Odds ratio
-0.381
0.026
0.107
0.031
0.00
0.40
0.68
1.03
-0.004
0.134
0.084
-0.080
0.194
0.082
0.072
0.071
0.066
0.074
0.96
0.06
0.24
0.22
0.01
1.00
1.14
1.09
0.92
1.21
-0.132
0.081
0.236
0.100
0.079
0.088
0.19
0.31
0.01
0.88
1.08
1.27
-0.011
-0.078
-0.041
0.111
0.168
0.250
0.152
0.099
0.101
0.107
0.110
0.104
0.106
0.098
0.91
0.44
0.70
0.31
0.11
0.02
0.12
0.99
0.93
0.96
1.12
1.18
1.28
1.16
0.069
0.056
0.22
1.07
-0.065
0.059
0.27
0.94
-0.196
-0.336
0.260
0.542
-0.686
0.064
0.051
0.096
0.090
0.151
0.00
0.00
0.01
0.00
0.00
0.82
0.71
1.30
1.72
0.50
1. Excludes households who primarily drink bottled water.
Note: Coefficients that are bold are statistically significant at the 10% level of significance. See Box 3 for methodology.
Source: Statistics Canada, Households and the Environment Survey (HES), 2006.
Statistics Canada – Catalogue no. 21-006-X
27
Rural and Small Town Canada Analysis Bulletin, Vol. 7, No. 5
Appendix Table A.7 Predicted percent of households which treat their drinking water1, showing
results for selected case households
Census metropolitan area
Case 1: Typical household
Census
agglomeration
Vancouver
Toronto
Montreal
Other
49.7
56.7
27.7
43.2
38.5
35.2
47.4
54.4
25.9
41.0
36.4
33.2
57.6
64.3
34.5
51.1
46.2
42.8
63.4
69.7
40.2
57.2
52.3
48.9
55.8
51.4
32.9
49.3
44.4
41.0
Rural
(2 persons aged 45-54; 1 person age 16-24; the
household member with the highest educational
attainment is a high school graduate (no postsecondary); income $50K-59K; water from a
municipal source; only tap water)
Case 2: Single
(1 person age 45-54, some post-secondary education;
income $30K-39K; water from a private source; only
tap water)
Case 3: Elderly couple
(2 persons age 55+; the household member with the
highest educational attainment has some post –
secondary education; income $60K-79K; water from
a municipal source; combination tap and other
drinking water)
Case 4: Mixed age, high education
(1 person age 16-24; 1 person age 45-54; 1 person
age 55+; the household member with highest
educational attainment has a university degree;
income $50K-59K; water from a municipal source;
tap water only)
Case 5: Young professional
(1 person age 25-44; university degree; income
$100K+; water from a municipal source; combination
tap and other drinking water)
1. Excludes households who primarily drink bottled water.
Note: See Box 3 for an explanation of the computation of predicted probability.
Source: Statistics Canada, Households and the Environment Survey (HES), 2006.
28
Statistics Canada – Catalogue no. 21-006-X
Rural and Small Town Canada Analysis Bulletin, Vol. 7, No. 5
Box 3 Methodology
A standard linear regression model was used to investigate the effect of water metering on total water use (see
Appendix A.3 and Appendix A.4). The model is based on Municipal Water and Wastewater Survey (MWWS)
2004 data. The dependent variable is the average daily domestic flow. The explanatory variables include metering
and location variables. Metering is measured by three dummy variables: high metering; some metering; and low
metering. The “low metering” category was used as the reference group and therefore dropped from the model.
The results were shown as relative to this reference group. Location variables include population density, Census
Metropolitan Area (CMA), Census Agglomeration (CA), strong Metropolitan Influenced Zone (MIZ), moderate
MIZ, and weak/no MIZ. As the reference group, “CMA” was dropped from the model and the results were
presented as relative to this group. All variables were dummy variables except for population density (for
definitions, please see Box 2).
Two alternative specifications were used. In the first specification, the average daily domestic flow is determined
by metering variables and population density only. In the second specification, we add also the regional dummies.
Each specification was estimated with two samples. The first was the full sample of 963 observations, for which
data was available, and the second was an “omitted outliers” sample that excluded the top and bottom 5% of
observations. A severe cut-off (removing the bottom 5% and top 5% of observations) was employed in order to
assess the sensitivity of the results to unusually high and low water use per capita in the MWWS 2004 dataset.
It should be noted that the analysis conducted with the MWWS 2004 was also replicated with Municipal Water Use
(MUD) survey data for 1998 and 2001, for the municipalities for which the same data was available. The results for
these previous years are similar to those obtained for 2004, with the main difference being that locational factors
(and particularly population density) were even less important in explaining water use levels when 1998 or 2001
data are used. In part, these differences can be explained by the inclusion of some rural communities in the 2004
database and a 2004 usable sample with lower average population density. Results for the MUD 1998 and MUD
2001 estimations are available from the authors upon request.
A logistic regression model was used to explain the household decision to treat drinking water. The dependent
variable of this model is the dichotomous choice “treat/do not treat” drinking water, coded as 1 and 0 respectively.
The explanatory variables included in the model are family size, age cohorts of household members, highest level
of education achieved by any household member, total household income, household’s type of water system
(municipal or private; private includes surface sources), type of water primarily drunk in the household,
municipality type (Box 2), and large urban indicators for Vancouver, Toronto, and Montreal. Each variable is
classified as a dummy variable (1 for true, 0 for false) except for family size, where we recorded the number of
family members in each of 5 age groups.
The dichotomous nature of treatment choices makes the analysis suitable for qualitative dependent variable
modeling (Long and Freese, 2001). Generally, this specification postulates the existence of a latent model, which is
continuous in its dependent variable (for instance, willingness to treat the water) but which is not observable in
reality. The model for the observable dichotomous outcome is derived from this latent process (see Long and
Freese, 2001). Hence, for the logistic regression, which is one form of qualitative dependent variable modeling, the
resulting specification is as follows,
Pr( y = 1 | x) =
exp(β0 + β1 x1 + ... + βn xn )
1
=
(
)
1 + exp β0 + β1 x1 + ... + βn xn
1 + exp
Statistics Canada – Catalogue no. 21-006-X
xβ
= Λ( xβ )
29
Rural and Small Town Canada Analysis Bulletin, Vol. 7, No. 5
Box 3 Methodology (continued)
This implies that the probability of observing a positive outcome (y=1), that is, the presence of water treatment, is a
function of a set of explanatory variables (x) defined by the logistic cumulative distribution function and Λ(.). In
this equation, the β’s represent the parameters to be estimated. Thus, using a more explicit notation for the set of
explanatory variables included in the model (demographic, socio-economic, source, and location) we can write the
model as,
Pr(Treat = 1 | x) = Λ( β1 Demo + β2 SocioEco + β3 Source + β4 Loc)
This equation represents the logistic regression model used in this study, which is estimated by maximum
likelihood methods and using the bootstrapping procedure in SPSS.
Bootstrapping. The sampling designs for Statistics Canada’s surveys are generally complex. As a result, the
variance of an indicator cannot be estimated with simple formulas. Therefore, re-sampling methods are often used
to estimate the variance. For the HES 2006 data, we use bootstrapping methods to estimate the variance of a
variable and to derive coefficients of variation as quality indicators of the estimates. Similarly we use bootstrapping
methods to make inferences in the logistic model. The bootstrap method consists of sub-sampling the initial sample
and then estimating the variance using the sub-sample results. For this computation we used bootstrap weights,
generated in the survey process. All these estimates were conducted using the BOOTVARE_V30.SPS program
(see Estimation of the Variance Using Bootstrap Weights User’s Guide for the BOOTVARE_V30.SPS Program
(Version 3.0), Statistics Canada, unpublished document).
Interpreting odds ratios. The odds ratio measures the effect of a unit change in the explanatory variable on the
ß
odds of observing a positive outcome (water treatment, in our case). The Odds Ratio is calculated as: OR = e k ,
where e is the base of the natural log and βk is the estimated coefficient for the kth variable. The exponential of the
coefficient is interpreted as: for a unit change in xk, the odds of observing a positive outcome (treatment) is
expected to change by a factor of “exp(ßk)”, holding all other variables constant. An odds ratio of 1 indicates the
explanatory variable has no effect on the treatment choice. For dummy variables, the odds ratio indicates the
change in odds as compared to the omitted category. For example, a household income of $80,000 to 90,000
increases the odds of treating water by a factor of 1.28, as compared to households with income less than $20,000
(Appendix Table A.6).
Predicted probabilities. Generally, the information provided by the logit coefficient on the relationship between
explanatory variables and outcomes is limited to the sign and statistical significance. A more meaningful
interpretation of this relationship comes from the computation of predicted probabilities (Long and Freese, 2001).
The predicted probability is the probability computed using the coefficient estimated by the model for any specific
value of the explanatory variables. They are computed as:
)
)
)
)
)
P(Treat = 1 | x) = Λ( β1 Demo + β 2 SocioEco + β 3 Source + β 4 Loc)
where the β-hat coefficients are those estimated with the logit model and the value of the explanatory variables
(Demo, SocioEco, Source and Loc) defines a specific household profile.
30
Statistics Canada – Catalogue no. 21-006-X
Rural and Small Town Canada Analysis Bulletin, Vol. 7, No. 5
Rural and Small Town Canada Analysis Bulletins (Catalogue no. 21-006-X)
Our latest editions
Vol. 7 No. 8: Seniors in Rural Canada
Kimberley Dandy and Ray D. Bollman
Vol. 7 No. 7: Structure and Change in Canada’s Rural Demography: An Update to 2006
Ray D. Bollman and Heather A. Clemenson
Vol. 7 No. 6: Rural Commuting: Its Relevance to Rural and Urban Labour Market
Spencer Harris, Alessandro Alasia and Ray D. Bollman
Vol. 7 No. 4: A Comparison of Rural and Urban Workers Living in Low Income
Myriam Fortin
Complete list of bulletins by major subject (note that some bulletins appear in more than one category)
Rural overview
Volume 1 No. 6; Volume 3 No. 3; Volume 4 No. 7; Volume 5 No. 2; Volume 6 No. 7
Demographics and
migration
Volume 1 No. 1; Volume 2 No. 2; Volume 2 No. 3; Volume 3 No. 6; Volume 4 No. 2; Volume 5 No.
4; Volume 6 No. 3
Education and skills
Volume 4 No. 5; Volume 5 No. 6; Volume 6 No. 2; Volume 7 No. 1
Agriculture
Volume 3 No. 2; Volume 4 No. 8; Volume 6 No. 1
Workforce and
employment
Volume 1 No. 2; Volume 2 No. 1; Volume 2 No. 6; Volume 2 No. 7; Volume 2 No. 8; Volume 3
No. 1; Volume 3 No. 4; Volume 3 No. 8; Volume 4 No. 1; Volume 4 No. 3; Volume 4 No. 7;
Volume 5 No. 5; Volume 6 No. 8
Business
Volume 1 No. 3
Tourism
Volume 5 No. 8; Volume 6 No. 5
Income and
expenditure
Volume 1 No. 4; Volume 2 No. 5; Volume 3 No. 7; Volume 4 No. 4; Volume 5 No. 7; Volume 7
No. 4
Housing
Volume 2 No. 4
Health
Volume 1 No. 5; Volume 4 No. 6; Volume 5 No. 3
Internet and
computer use
Volume 1 No. 7; Volume 3 No. 5; Volume 5 No. 1; Volume 7 No. 3
Social trends
Volume 6 No. 4; Volume 7 No. 1
Environment
Volume 6 No. 6; Volume 7 No. 2
Aboriginal and the
north
Volume 1 No. 8
Statistics Canada – Catalogue no. 21-006-X
31